Форум » Философия хай-энда » Про искажения -- в том числе и музыкальные » Ответить
Про искажения -- в том числе и музыкальные
Wavebourn: Вот моя статья, писал для чайников на международном форуме, пытался донести максимально просто и понятно. Может, кому пригодится... Some people argue that the industry goes for less and less distortions, but people don't like clean sound. There is a well established belief that tubes and transformers produce "good" distortions. There are no "good" distortions! Some of them are more audible, some are less. For example, vintage class A tube amps had the more distortions the higher the power, that's why they were measured for percent of distortions per power, or for higher power with less percent of distortions. It was a comparison of apples to apples. When push-pull amps were made they allowed to jump to the next step in the game: being more efficient they produced more power per the same percentage of distortions. It was a comparison of apples to oranges, since push-pull amps had less of even-order distortions, so measurements were better even when they sounded the same or worse. Transistor amps could not work without deep negative feedbacks that linearized transistor stages. As the result of feedback overall percentage of distortions on higher power was very low. It was easy for them to generate less of harmonic distortions' percentage on higher power! It was a comparison of apples to shrimps... Transistor amps were optimized for higher output per consumed from outlet power, in order to achieve this criterion they were made such a way, the lower is power, the higher is level of distortions, the wider is specter... But the higher is sound pressure level, the more forgiving is our perception to non-linear distortions. Reason: all physical media distort, including one of which our ears are made. "Natural" distortions add perception of higher volume, it is hard to tell what distorts, the amp or the air (or the ear!). But however the less of them is added by electronics, the more are allowed for an air and ears. When oscillating physical media distort sounds the higher is it's deformation the higher is pitch of distortions. When you kick a piano string by it's hammer hard you hear higher harmonics, if to speak language of Fourier transforms. If you play soft you hear lower order of harmonics. When the natural sound decay higher harmonics decay faster. So, when an amp generates higher harmonics on higher power our perception forgives that. But when on lower power percentage of distortions goes up and specter goes wider it is unnatural, contrary to expectations, and such an amp sounds less natural. So, the right amp have to distort the less the less is the power. But how many of them are forgivable on higher power? As less as possible. You probably can't hear distortions of some vintage class A amp on higher power on a saxophone, or a human voice solo; it will be perceived like harmonics of the same instrument, but an ensemble or a chord of one instrument have more than one note playing simultaneously, so inter-modulating they will produce harsh sounds that did not present originally. There are no such devices that don't distort. The ideal device should have a transfer function Y=X*K where X - input signal, K - amplification coefficient. We can imagine an exponential transfer function as a sum of functions of different orders, like Y=K1*X + K2*X^2 + K3*X^3 +K4*X^4, and so on. Very short part of this function may be seen as a linear one; the finer are chunks, the closer each one seems to be to a linear function. That means, we have to take a real device with higher gain than needed, but use a small part of it's characteristic. No other ways exist. Another problem is, each device need time to react. What is worse, speed of reaction is modulated by the signal itself. Theoretically is seems simple: 1. Take an amp with fewer tubes, enormous power capabilities with less power losses from stage to stage, but use a little bit of it's power amplification capabilities. It is a very expensive way, and it has a limit: the more of power is dissipated, the heavier are parts, the more reactive they are, the more noises and frequency dependencies your get. 2. Take an amp with enormous gain, add feedback, and it's curve is straightened. The problem is, a first, low order function multiplied by itself gives higher order. Lowering distortions that our perception can forgive we are adding what it forgives less. Second, because of the time needed for reaction for each amplification stage phase shifts are accumulated from stage to stage, and on some high frequency where phase shift is 180 degrees a negative feedback becomes a positive one forcing an amp to oscillate, if the overall gain through the loop on this frequency is slightly higher than one. Now, we need to compensate the amp to stop oscillations, for example adding higher losses on that frequency, but the same time slowing down it's reaction! There is the 3'rd way: linearization. For example, assisting a device with more devices that stretch it's transfer function. For example, a triode may be loaded on a constant current source. Or a source follower bootstrapped. And, finally, the 4'th way: use optimal combination of all 3, optimizing such a way that most audible distortions will be mostly minimized. No amps with no distortions can be made. Add here speakers that are driven by amps, so the system containing an amp and speakers have to be optimized together, otherwise a perfect by itself amp with a perfect by itself speaker in combination produce worse sound than less expensive combination that had been optimized all together. Does the industry go this way? I can't tell for the all industry, only for myself...
Ответов - 39, стр:
1 2 All
полная версия страницы